Nvidia has committed roughly $40 billion to artificial intelligence companies in 2026, with an estimated $30 billion flowing to a single recipient: OpenAI. The chipmaker’s latest move, a $500 million equity-linked deal with fiber optics giant Corning announced in May 2026, is modest by comparison. But taken together, the investments have triggered a pointed debate on Wall Street: Is Nvidia bankrolling its own customers so they turn around and buy more Nvidia chips?
How the numbers add up
The $30 billion OpenAI commitment is the centerpiece. Multiple outlets, including Bloomberg and Reuters, reported that Nvidia invested approximately $30 billion as part of OpenAI’s record $40 billion funding round, which closed in early April 2026. Beyond OpenAI, Nvidia has taken equity positions in infrastructure and AI companies including Corning, and has been linked to investments in firms like xAI and CoreWeave, pushing the total past the $40 billion mark according to tallies by financial analysts tracking the deals.
Nvidia ended its most recent fiscal quarter with more than $43 billion in cash and short-term investments, meaning the company has the balance sheet to absorb this level of spending. But the sheer concentration of capital flowing to companies that are also Nvidia’s biggest hardware buyers is what has analysts raising flags.
The Corning deal, up close
The Corning partnership offers the clearest window into how these arrangements work, because Corning is publicly traded and subject to SEC disclosure rules. According to a Corning regulatory filing, the deal will increase U.S. optical connectivity manufacturing capacity tenfold, boost fiber capacity by more than 50%, open three new domestic manufacturing facilities, and create over 3,000 jobs. Bloomberg reported that Nvidia purchased $500 million in rights for Corning shares, tying Nvidia’s financial upside directly to Corning’s expansion in AI data center infrastructure.
Nvidia is not simply selling GPUs and walking away. It is taking an equity stake in a company whose growth depends on building the physical networks that connect Nvidia-powered servers. If Corning thrives, Nvidia profits twice: once from the chips inside the data centers and again from its ownership stake in the company wiring those data centers together.
The OpenAI question
If the Corning deal is a window, the OpenAI investment is the whole building. Alongside the roughly $30 billion financial commitment, the two companies announced a strategic partnership to deploy what they described as 10 gigawatts of Nvidia systems. That figure represents the total rated capacity of the planned deployments, roughly equivalent to the peak output of ten large nuclear power plants, though actual simultaneous power draw would depend on workload and utilization.
OpenAI is simultaneously one of Nvidia’s largest customers and one of its largest investment recipients. The exact terms of Nvidia’s stake, whether structured as direct equity, convertible instruments, warrants, or some combination, have not been disclosed in a public regulatory filing as of late May 2026. Nor is it clear whether the investment carries any contractual obligation for OpenAI to purchase Nvidia hardware specifically. But the financial loop is hard to miss: billions flow from Nvidia to OpenAI, and billions flow back as OpenAI builds out compute infrastructure dominated by Nvidia GPUs.
Why “circular deal” is more than a talking point
The criticism is not coming from the fringes. New Street Research analyst Pierre Ferragu has publicly questioned whether Nvidia’s investments in AI companies create a feedback loop that inflates both Nvidia’s revenue and the apparent health of the broader AI sector. In a May 2026 research note distributed to New Street clients, Ferragu raised concerns that the investment-and-purchase cycle could obscure how much of Nvidia’s revenue growth reflects independent end-customer demand versus capital that Nvidia itself supplied. The logic follows a straightforward path: Nvidia writes a check to an AI company, that company spends a large share of the capital on Nvidia hardware, and Nvidia books the resulting sales as organic revenue growth. The same dollars get counted twice, once as an investment outflow and again as top-line revenue.
If a meaningful portion of AI infrastructure spending is ultimately financed by Nvidia itself, then the growth rates that have propelled Nvidia’s stock past a $3 trillion market capitalization may overstate how much independent, end-customer demand actually exists. For investors trying to value Nvidia, the distinction between “customers spending their own money” and “customers spending Nvidia’s money” is not academic. It goes to the heart of whether current revenue trends are sustainable or partially self-generated.
The Cisco precedent
Tech veterans have heard a version of this story before. In the late 1990s, Cisco Systems ran an aggressive vendor financing program, extending billions in loans and credit to telecom companies that used the funds to buy Cisco networking equipment. Revenue soared. Then the dot-com bubble burst, many of those customers defaulted, and Cisco was left writing off billions in bad debt. The company’s stock, which peaked near $80 in March 2000, fell below $9 by late 2002 and has never returned to its inflation-adjusted highs.
The parallel is imperfect, and Nvidia’s defenders are quick to say so. Nvidia is taking equity positions rather than extending loans, which means it shares in upside rather than simply bearing credit risk. The AI companies receiving Nvidia’s capital, particularly OpenAI with its reported $10 billion-plus in annualized revenue, have business fundamentals that dwarf the speculative telecoms of the late 1990s. But the structural similarity is real: a dominant hardware supplier is financing the customers who buy its products, and the resulting revenue growth is at least partially self-generated. Whether this cycle ends differently depends on whether AI demand proves durable enough to justify the spending, or whether the money is chasing a feedback loop.
What Nvidia and its partners have said
Nvidia has not directly addressed the circular-deal framing in official statements. CEO Jensen Huang’s public remarks have focused on the scale of AI infrastructure demand and the need for ecosystem investment to keep pace. The Corning partnership announcement emphasized U.S. manufacturing, job creation, and the physical infrastructure required to support AI workloads. The OpenAI partnership announcement centered on the 10-gigawatt deployment target and the technical demands of next-generation AI models.
Supporters of Nvidia’s approach argue that strategic investments in ecosystem partners are standard practice in capital-intensive technology markets. Intel Capital spent decades investing in companies that built products around Intel processors. Qualcomm’s venture arm has done the same in mobile. From this perspective, Nvidia is simply operating at a larger scale because the AI infrastructure buildout demands larger capital commitments. The investments, in this reading, are not financial engineering but a rational strategy to ensure that the supply chain, from silicon to fiber optics, can keep pace with demand.
Notably, AMD, Nvidia’s closest competitor in AI accelerators, has not pursued equity investments in customers at anything close to this scale. That asymmetry is part of what makes the Nvidia strategy so unusual and so difficult to evaluate from the outside.
What earnings reports and regulatory filings will reveal
Several open questions will determine whether the circular-deal critique gains regulatory traction or fades into background noise.
First, Nvidia’s quarterly earnings filings will eventually need to disclose how much revenue comes from companies in which Nvidia holds significant equity positions. Related-party transaction disclosures, required under SEC rules, could clarify the degree of financial entanglement. Investors should pay close attention to Nvidia’s next 10-Q filing for any new related-party footnotes.
Second, the Federal Trade Commission and Department of Justice have signaled heightened interest in AI market concentration. If regulators conclude that Nvidia’s dual role as supplier and financier distorts competition or creates barriers to entry for rival chipmakers, the investments could face antitrust scrutiny.
Third, and perhaps most important, is the question of independent demand. If AI companies that received Nvidia capital continue to grow revenue from paying customers, the circular-deal critique loses force. The investments would look, in hindsight, like smart bets on companies that were going to buy Nvidia hardware regardless. But if growth stalls and these companies struggle to generate returns without continued infusions of Nvidia capital, the feedback loop will look far more fragile.
The verified facts point in one direction: Nvidia is becoming both the supplier and the financier of the AI boom, embedding itself across the value chain from silicon to fiber optics. Whether that makes Nvidia the architect of a durable industrial ecosystem or the engineer of a self-referencing growth cycle will be determined by earnings reports, regulatory filings, and whether the companies Nvidia is funding can eventually stand on their own.
